A backoff approach to design of optimally flexible water networks under uncertainty

Abstract

Water management through design of proper water networks is a key to the reduction of freshwater usage and wastewater discharge, and thus, more sustainable production in process industries. In this work, an optimization-based approach to synthesis of water networks under uncertainties (e.g., disturbances) in both contaminants concentrations and flow rates is presented. A backoff procedure from the literature is embedded in the optimization problem in order to guarantee steady-state feasibility of the optimally designed network in spite of the sustained process disturbances. Appropriate reformulations for improving mathematical properties of the problem result in a mixed-integer nonlinear program and enables its global solution using deterministic algorithms. The network structure features tanks with constant-flow pumps and overflow lines for better disturbance handling with little to no need for an active control system, thereby reducing network’s operational complexity. The proposed approach is demonstrated through a number of case studies including a real corn refinery problem with multiple contaminants, where it is shown that an optimally flexible water network can be achieved with a minimal increase (only 6%) in the freshwater consumption compared with a nominally optimal design.